def test_pca_invalid():
X = pd.DataFrame([[3, 0, 1, 6],
[1, None, 1, 6],
[10, 2, 1, 6],
[10, 2, 2, np.nan],
[None, 2, 2, 5]])
pca = PCA()
with pytest.raises(ValueError, match="must be all numeric"):
pca.fit(X)
X = pd.DataFrame([[3, 0, 1, 6],
['a', 'b', 'a', 'b'],
[10, 2, 1, 6],
[10, 2, 2, 23],
[0, 2, 2, 5]])
pca = PCA()
with pytest.raises(ValueError, match="must be all numeric"):
pca.fit_transform(X)
X_ok = pd.DataFrame([[3, 0, 1, 6],
[1, 2, 1, 6],
[10, 2, 1, 6],
[10, 2, 2, 5],
[6, 2, 2, 5]])
pca = PCA()
pca.fit(X_ok)
with pytest.raises(ValueError, match="must be all numeric"):
pca.transform(X)
def test_pca_array():
X = np.array([[3, 0, 1, 6], [1, 2, 1, 6], [10, 2, 1, 6], [10, 2, 2, 5],
[6, 2, 2, 5]])
pca = PCA()
expected_X_t = pd.DataFrame(
[[3.176246, 1.282616], [4.969987, -0.702976], [-3.954182, 0.429071],
[-4.079174, -0.252790], [-0.112877, -0.755922]],
columns=[f"component_{i}" for i in range(2)])
pca.fit(X)
X_t = pca.transform(X)
assert_frame_equal(expected_X_t, X_t.to_dataframe())
def test_pca_woodwork_custom_overrides_returned_by_components(X_df):
y = pd.Series([1, 2, 1])
override_types = [Integer, Double]
for logical_type in override_types:
X = ww.DataTable(X_df, logical_types={0: logical_type})
pca = PCA(n_components=1)
pca.fit(X)
transformed = pca.transform(X, y)
assert isinstance(transformed, ww.DataTable)
assert transformed.logical_types == {
'component_0': ww.logical_types.Double
}